Septin‐regulated actin dynamics promote Salmonella invasion of host cells

Abstract Actin nucleators and their binding partners play crucial roles during Salmonella invasion, but how these factors are dynamically coordinated remains unclear. Here, we show that septins, a conserved family of GTP binding proteins, play a role during the early stages of Salmonella invasion. We demonstrate that septins are rapidly enriched at sites of bacterial entry and contribute to the morphology of invasion ruffles. We found that SEPTIN2, SEPTIN7, and SEPTIN9 are required for efficient bacterial invasion. Septins contributed to the recruitment of ROCK2 kinase during Salmonella invasion, and the downstream activation of the actin nucleating protein FHOD1. In contrast, activation of the ROCK2 substrate myosin II, which is known to be required for Salmonella enterica serovar Typhimurium invasion, did not require septins. Collectively, our studies provide new insight into the mechanisms involved in Salmonella invasion of host cells.

Septin recruitment was observed in approximately 40%-60% of invasion sites (Figure 1b), possibly reflecting the dynamic nature of actin polymerisation at this site.
For high-resolution visualisation of actin and septins during S. Typhimurium invasion, HeLa cells were infected with S. Typhimurium for 10 min, and invasion ruffles were examined by structured illumination microscopy (SIM; Figure 1c). For these experiments the plasma membrane was labelled with wheat germ agglutinin (WGA), which binds to cell surface N-acetylglucosamine and N-acetylneuraminic acid residues (Wright, 1984). Following 3D reconstruction, actin was found to occupy filamentous structures projecting dorsally from the cell surface in association with WGA+ plasma membrane ( Figure 1d). Endogenous SEPTIN7 associated with these actin filaments in adjacent microdomains. These findings are consistent with prior studies indicating a role for septins in serving as a scaffold for F-actin filaments.

| Septins are required for efficient S. Typhimurium invasion of host cells
To determine whether the close association of septins with actin within the S. Typhimurium invasion ruffles affects bacterial internalisation, HeLa cells were treated with siRNA targeting SEPTIN2, SEPTIN7, or SEPTIN9. Depletion of septins was confirmed by Western blot analysis ( Figure S1a,b). Cells were subsequently infected with S. Typhimurium for 30 min. In these experiments, we used a longer infection time to enrich for infected cells and to easily discern internalised bacteria.
Immunostaining before permeabilisation was used to differentiate between intracellular and extracellular bacteria, as previously described by Smith et al. (2007).
Knockdown of SEPTIN2, SEPTIN7, and SEPTIN9 individually, caused a significant decrease in S. Typhimurium invasion relative to control siRNA-treated cells (Figure 2a). An invasion defect was also observed when treating HeLa and Henle 407 cells with SEPTIN7 siRNA pools ( Figure S2a,b). Each siRNA pool contained two independent siRNAs targeting SEPTIN7 and knockdown efficiency was confirmed ( Figure S2c-f). Together, these results demonstrate a role for septins during the initial stages of S. Typhimurium infection.
SEPTIN7 is the stabilising septin within its hetero-oligomeric complex (Fung et al., 2014). SEPTIN7 siRNA treatment not only reduces levels of SEPTIN7 in the cell, but it also causes the destabilisation of other septin isoforms, unlike SEPTIN2 and SEPTIN9 siRNA ( Figure   S1b). For this reason, we employed SEPTIN7 siRNA as a tool for subsequent studies of septin function during infection.
Since septin-depleted cells have a significant bacterial internalisation defect, we examined the effect of septin knockdown on the morphology of invasion ruffles. Scanning electron microscopy (SEM) was used to obtain high-resolution images of S. Typhimurium invasion sites (Figure 2b). HeLa cells were treated with the indicated siRNA and infected with S. Typhimurium for 10 min. Control siRNAtreated cells displayed large membrane protrusions including fingerlike projections, known as filopodia, at the leading edge of the invasion ruffles. The cell surface also appeared rough with small filopodia-like structures, consistent with previous studies (Truong et al., 2013). In contrast, knockdown of SEPTIN7 resulted in invasion ruffles with few filopodia, and the cell surface appeared smooth. Knockdown of SEPTIN7 also led to the formation of large sheet-like protrusions (lamellipodia) that extended across the cell surface.
We also used live cell imaging to visualise the impact of SEPTIN7

| Septins promote ROCK2 recruitment to S. Typhimurium invasion sites
Previous studies have demonstrated that septins function as a molecular platform to bring myosin and myosin kinases in close proximity for maximal myosin activation during contractile processes (Joo et al., 2007). RhoA-associated kinases (ROCKs) are well-studied activators of myosin II-mediated contractility (Wilkinson, Paterson, & Marshall, 2005). ROCKs can directly phosphorylate myosin light chain II (MLC2) on Ser19 and Thr18 to promote myosin II activation, therefore promoting actomyosin contractility. This pathway is important for the RhoA-myosin II-dependent but Arp2/3 independent pathway, of S. Typhimurium invasion (Hanisch, Kolm, Wozniczka, Bumann, & Rottner, 2011). Interestingly, ROCK2 has been specifically implicated in the S. Typhimurium invasion process through its role in FIGURE 1 Septins are recruited to Salmonella enterica serovar Typhimurium invasion sites. Septin recruitment to the invasion ruffle was assessed in HeLa cells. (a) HeLa cells were infected with S. Typhimurium and fixed 10-min postinvasion. Cells were then immunostained for endogenous SEPTIN2, SEPTIN7, SEPTIN9, and SEPTIN11 (green); F-actin (red); and S. Typhimurium (blue). Scale bar, 11 μm. Images were taken using a spinning-disk confocal microscope and invasion site is indicated by an arrow. (b) Quantification of septin recruitment to the invasion ruffle was done in 100 infected cells in three independent experiments. (c) Septin domain localisation to the invasion ruffle was assessed in HeLa cells using structured illumination microscopy. HeLa cells were infected with S. Typhimurium and fixed 10-min postinvasion. Cells were then immunostained for endogenous SEPTIN7 (green), F-actin (red), and the membrane marker wheat germ agglutinin WGA (blue). (d) 3D reconstruction of the membrane protrusion during invasion was created using image Z stacks. Area examined using 3D reconstruction is indicated by arrow phosphorylating and activating FHOD1, an actin nucleator that promotes filopodia production (Truong et al., 2013). Thus, we examined the effect of SEPTIN7 knockdown on localisation of ROCK2 to the invasion ruffle during S. Typhimurium invasion.
HeLa cells were treated with SEPTIN7 siRNA 48 hr prior to infection. Subsequently, cells were infected with S. Typhimurium for 10 min. Endogenous ROCK2 colocalised with actin-enriched invasion sites in control siRNA treated cells (Figure 3a, upper panels), an observation consistent with previous findings (Truong et al., 2013). Following SEPTIN7 knockdown, we observed a decrease in the frequency of ROCK2 recruitment to the invasion ruffle (Figure 3a,b). Therefore, septins contribute to ROCK2 recruitment to S. Typhimurium invasion sites. Impairment of ROCK2 recruitment to invasion sites could affect ROCK2-dependent activation of myosin II and FHOD1, which are ROCK2 substrates known to play roles in S. Typhimurium invasion (Hanisch et al., 2011;Truong et al., 2013). Thus, we examined the effect of SEPTIN7 knockdown on activation of ROCK2 substrates during S. Typhimurium invasion.

| Myosin II recruitment and activation during S. Typhimurium invasion does not require septins
Nonmuscle myosin II is a major contributor to cellular organisation and  Post-siRNA transfection, cells were infected with S. Typhimurium for 30 min. Differential antibody staining was used to identify intracellular and extracellular bacteria. 100 cells were analysed for bacterial infection. Data is normalised to cells treated with control siRNA. "*" denotes p value < 0.05. (b) Scanning electron microscopy of S. Typhimurium invasion sites upon septin knockdown. HeLa cells were transfected with the indicated siRNA and subsequently infected with S. Typhimurium for 10 min. Cells were then fixed with gluteraldehyde, and scanning electron microscopy images were taken at 8000x g. Scale bar, 5 μm. (c) Live cell imaging of S. Typhimurium invasion sites upon septin knockdown. HeLa cells were transfected with indicated siRNA. 24-hr post-siRNA knockdown, cells were transfected with LifeAct-mRFP (shown in green) and then infected 24 hr later with wild -type S. Typhimurium. Invasion was recorded in live cells using spinning-disk confocal microscopy. Times shown (in minutes) are relative to the initiation of the invasion process. Using Volocity analysis system, 3D reconstruction was produced to examine invasion ruffle formation under control and septin siRNA treated cells Myosin II was demonstrated to localize to S. Typhimurium invasion sites where it contributes to internalization of the bacteria. (Hanisch et al., 2011). Phosphorylation is required for Myosin II activity during contractile actions and it is known that myosin II phosphorylation occurs near sites where septin filaments are associated with actin stress fibres (Joo et al., 2007). Since septins can bind to septinassociated Rho guanine nucleotide exchange factor (SA-Rho-GEF) and myosin, a signalling cascade of SA-Rho-GEF-RhoA-ROCK-myosin II, which is essential for complete myosin II activation and thus myosin-actin interaction, could be enabled by septin scaffolding (Nagata & Inagaki, 2005). Thus, it is possible that septins contribute to the localisation or activation of myosin II during S. Typhimurium invasion.
To examine whether myosin II requires septins for localisation to the invasion ruffle, we tested the recruitment of the myosin II heavy

| Septins promote FHOD1 phosphorylation during S. Typhimurium invasion
FHOD1, a formin family member and host cell actin nucleating protein, has been shown to drive filopodia formation during the initial stage of S. Typhimurium invasion (Truong et al., 2013). Moreover, ROCK2 specifically mediates FHOD1 phosphorylation and activation during S. Typhimurium invasion (Truong et al., 2013). Since SEPTIN7 knockdown decreased recruitment of ROCK2 to S. Typhimurium invasion sites and we observed a striking loss of filopodia in invasion ruffles of SEPTIN7 knockdown cells, we tested whether septins act as a molecular scaffold for FHOD1 recruitment and activation as well. To test this, we examined the FHOD1 phosphorylation state following SEPTIN7    (Finlay et al., 1991). Live cell imaging has revealed that the early steps of invasion include rapid extension of F-actin rich filopodia, followed by generation of lamellipodial structures and contraction of these cell surface protrusive structures (Hanisch et al., 2011;Truong et al., 2013). Subsequently, the plasma membrane is invaginated, and bacteria are internalised into cells in Salmonella-containing vacuoles (Finlay et al., 1991;Terebiznik et al., 2002). In our study, we observed rapid enrichment of septins at S. Typhimurium invasion sites and their association with F-actin rich microdomains. Knockdown of SEPTIN7 was used as a tool to disrupt the septin cytoskeleton, since it is essential for stability of other septins (Tooley et al., 2009). Under these conditions, we observed dramatic alterations to invasion sites, including a loss of filopodia and disorganisation of the lamellipodial protrusions.
Our data are consistent with septins serving as a scaffold for the organisation of polymerised actin. It is interesting to note that although SEPTIN7 knockdown disrupts the septin cytoskeleton, we observed that SEPTIN2 and 9 could still be recruited to S. Typhimurium invasion sites in SEPTIN7 knockdown cells ( Figure S3). This is likely due to their ability to bind phospholipids such as phosphatidylinositol 4,5bisphosphate (PtdIns (4,5)2) (Zhang et al., 1999), which is temporally enriched in S. Typhimurium invasion ruffles.
Our data also suggests that septin molecular platforms promote signal transduction events at S. Typhimurium invasion sites. We find that ROCK2 recruitment to invasion sites and activation of its downstream target FHOD1 are impaired in SEPTIN7 knockdown cells.
Although septin depleted cells were shown previously to be impaired in generating PP-MLC mediated by ROCK2 (Joo et al., 2007) we did not see an inhibition of myosin activation in SEPTIN7 depleted cells during invasion. This suggests that kinases other than ROCK2 may contribute to myosin II activation during S. Typhimurium invasion. In this regard, S. Typhimurium has been shown to induce elevation of intracellular calcium during invasion (Ruschkowski, Rosenshine, & Finlay, 1992), which may mediate myosin II activation through calcium-activated kinases. Also, the S. Typhimurium lipid phosphatase known as SopB has been shown to induce PP-MLC activation when transfected in HeLa cells (Wasylnka et al., 2008)

| Bacterial strains and infection
S. Typhimurium SL1344 was used in this study (Hoiseth & Stocker, 1981). Late-log bacterial cultures were used for infecting HeLa and Henle 407 cells during experiments as outlined previously (Szeto, Namolovan, Osborne, Coombes, & Brumell, 2009). Briefly, WT bacteria were grown for 16 hr at 37°C with shaking and then subcultured

| Live cell imaging
Cells were grown on 2.5-cm coverslips, transfected 12-16 hr before invasion with LifeAct-mRFP constructs and preincubated with RPMI-1640 media (supplemented with L-glutamine, HEPES, no bicarbonate; Wisent) with 10% FBS at 37°C for 20 min. Cells were infected with WT SL1344 bacteria. In brief, 1 ml of late log bacterial suspension was extensively washed with PBS. The bacterial suspension was used for infection. Time-lapse confocal z-stacks of the cells were imaged using a Leica DMI 6000B inverted fluorescence microscope with a Hamamatsu ImagEMx2 camera (Quorum Technologies Inc., Guelph, Canada). Images were processed using Volocity 6 software (PerkinElmer).

| Scanning electron microscopy
Cells were seeded in a 24-well tissue culture plate at a density of
Immunostaining before permeabilisation was used to differentiate between intracellular and extracellular bacteria (Smith et al., 2007).
Coverslips were mounted onto glass slides using DakoCytomation fluorescence mounting medium and imaged using a Leica DMIRE2 inverted epifluorescence microscope or a Quorum spinning disk confocal microscope (Leica DMI6000B inverted fluorescence microscope, Hamamatsu ORCA Flash 4 sCMOS and colour camera) and processed using Volocity 6 software (Perkin Elmer).
For Structured Illumination Microscopy (SIM), high precision glass coverslips with 1.5-mm thickness were mounted onto glass slides using Prolong Diamond Antifade Mountant and acquired on Zeiss Elyra PS1 equipped with an Axio Observer Z1 microscope, Andor iXon3 885 detectors and 60 × /1.4 NA plan-Apochromat oil immersion objectives. Z-stacks were collected and computationally deconvolved using Zeiss Zen 2012 with SIM licence.
Sample buffer (60 mM Tris pH 6.8, 5% glycerol, 1% SDS, 2% βmercaptoethanol, 0.02% bromophenol blue) was added to the suspension, and samples boiled for 6 min. Samples were separated on 8% SDS-PAGE gel for phospho-FHOD1 blots and 12% for septin blots, then they were transferred to Polyvinylidene Fluoride (PVDF) membranes. Membranes were blocked in 3% BSA or 5% milk in TBS-T overnight. Primary antibodies were incubated overnight at 4°C. Secondary antibodies used were conjugated to horseradish peroxidase (HRP) and were purchased from Sigma. Densitometry was performed on scanned immunoblot images using the ImageJ gel analysis tool (Abramoff, Magalhaes, & Ram, 2004).

| Statistical analysis
Statistical analyses were conducted using GraphPad Prism v5.0. The mean +/− standard error of the mean is shown in figures, and p values were calculated using one sample t test or one-way analysis of variance (ANOVA), where indicated. A p value of less than 0.05 was considered statistically significant and is denoted by *. p < 0.01 is denoted by ** and p < 0.001 is denoted by ***.